Home - Power Supply Knowledges - Constant Power Drivers for LED Grow Lights: Why Wattage Is the Wrong Number to Buy On
Constant Power Drivers for LED Grow Lights: Why Wattage Is the Wrong Number to Buy On
A grow-light fixture is usually sold on two numbers: the wattage on the box and the price per fixture. Neither of them tells you what the fixture will actually deliver in a room at 28 °C and 70% humidity after eight hours of operation, which is the only condition that matters for a crop.
The driver is where that gap comes from. A 600 W fixture with a constant current driver delivers something less than 600 W once the LED array reaches operating temperature, because the array’s forward voltage falls as it heats and a constant current driver holds current rather than power. A 600 W fixture with a constant power driver delivers 600 W at 25 °C and 600 W at 65 °C, which is why the second one costs more and why the first one is the reason growers think all LED fixtures are overrated.
This article explains the three regulation modes, the temperature mechanism that makes the difference, and the arithmetic that turns a 5% driver efficiency gap into an annual number a grower can actually evaluate.
Three modes, and what each one holds
Every LED driver regulates one of three quantities.
Constant voltage (CV) holds the output voltage. The current is whatever the load draws at that voltage, following Ohm’s law across the array. This is the mode used for LED strip, signage and most architectural lighting, where the load is a defined 24 V product.
Constant current (CC) holds the output current. The voltage floats to whatever the array needs. This is the standard for high-power LED strings, because LEDs are current-driven devices and light output tracks current far more predictably than it tracks voltage.
Constant power (CP) holds the product of the two. As the array’s forward voltage falls with temperature, the driver raises the current to compensate, and the wattage stays put. The output voltage is allowed to move within a window — typically 60% to 100% of the rated maximum — and below that window the driver reverts to constant current at its maximum.
For a grow light, the third mode exists for one reason: photosynthesis scales with photons, photons scale with electrical power (within a fixed efficiency), and the array’s forward voltage is not stable.
The temperature mechanism nobody accounts for
An LED die’s forward voltage falls by roughly 2 to 4 mV for every degree Celsius of junction temperature rise. That sounds negligible until you multiply it by the number of dies in series.
Take a 600 W fixture built as four parallel strings of 60 dies each, running at 3.0 V per die at 25 °C. The string voltage is 180 V. Run the fixture for an hour and the junction temperature climbs from 25 °C to 65 °C. At 3 mV/°C per die, each die loses 120 mV, so the string voltage drops to 172.8 V — a 4% fall.
Now consider what each mode does with that:
| Mode | What happens as Vf falls 4% | Effect on delivered power |
| Constant voltage | Current rises 4% (the array presents a lower resistance) | Power rises 4% |
| Constant current | Current held; voltage falls 4% | Power falls 4% |
| Constant power | Current rises 4% to compensate | Power held flat |
The constant voltage case is the dangerous one and the constant current case is the disappointing one. In CV mode, a fixture that was drawing 600 W cold draws 624 W hot, which raises the junction temperature further, which lowers Vf further. On a well-designed array with adequate heatsinking the loop settles. On a marginal one it does not, and the failure mode is a string of dead emitters in the middle of the board where the airflow was worst — the classic “burnt spot” that growers blame on the LEDs when the driver was holding voltage.
In CC mode the opposite happens: the fixture delivers 4% less power as it warms up, so the 600 W on the box is a cold-start number. Photon output drops proportionally. The grower, measuring with a PAR meter after a two-hour warm-up, gets 4% less than expected and concludes the fixture is overrated — which, in the terms that matter, it is.
Constant power is the mode that makes the number on the box true at operating temperature. That is the whole argument, and it is worth about 4% of your yield assumption.
What CP costs, and when it is not worth it
Constant power requires the driver to multiply output voltage by output current continuously and adjust the loop. In practice this is a small analogue multiplier in the feedback path, or a digital implementation in a microcontroller-based controller. It adds maybe $2–4 to a 600 W driver, and it constrains the output voltage window, because the multiplier has to stay within its linear range.
That constraint is where projects go wrong. If the array’s Vf range across the temperature span is wider than the driver’s CP window, the driver drops out of CP at one end and behaves like a CC supply. A fixture specified at 600 W might deliver 600 W at 40 °C and 570 W at 70 °C, and nobody notices because the datasheet only has one number.
So the specification should include the window: constant power maintained from 60% to 100% of rated output voltage, across the declared ta range. That sentence forces the supplier to test at both temperature extremes, which is the only way to know.
If your fixture has a narrow Vf range — for example, a fixed array with a well-controlled junction temperature and a driver sized tightly — constant current is fine and cheaper. CP earns its cost when the fixture operates across a wide ambient range, when the array’s thermal design is variable, or when the fixture is sold on a guaranteed wattage that a grower will check.
µmol/J is the number that actually decides profit
Growers are moving from watts per square foot to photons per joule, and the driver participates in that number directly.
Fixture efficiency is usually quoted as µmol/J — micromoles of photosynthetically active photons per joule of electrical input. Good bar-type fixtures in 2026 run 2.7 to 3.0 µmol/J; the best run around 3.1; cheap fixtures run 2.2 to 2.4. The difference between 2.7 and 3.1 is 15% more light for the same electricity, which is far larger than the difference between a good driver and a cheap one.
But the driver’s contribution is not zero, and it compounds. Here is the arithmetic for a 200-fixture room:
| Driver efficiency 88% | Driver efficiency 93% | |
| Fixture input | 400 W | 400 W |
| Power to LEDs | 352 W | 372 W |
| Photons (at 2.8 µmol/J LED efficiency) | 986 µmol/s | 1,042 µmol/s |
| Energy per fixture, 12 h/day, 365 days | 1,752 kWh | 1,752 kWh |
| Useful photons per year, 200 fixtures | 62.2 × 10⁹ mol | 65.7 × 10⁹ mol |
Same electricity bill, 5.5% more photons. If you would rather hold photons constant and save energy, the same 5% buys you 88 kWh per fixture per year, or 17,500 kWh across 200 fixtures — about $2,600 a year at $0.15/kWh, before any utility rebate.
That is the case for spending $6 more on a driver. It is also why the “5% efficiency” line on a datasheet is not a rounding error in this application: it is the difference between two fixtures that a grower will price identically.
One caution: the efficiency figure has to be measured at the operating point. A driver that is 93% efficient at 100% load is typically 89–90% at 50% load, and most growers dim to 60–70% for much of the cycle. Ask for the efficiency curve at 50%, 75% and 100%, not a single number.
Dimming, control and the flicker question
Horticultural fixtures are increasingly dimmed and scheduled. Three control methods dominate:
0-10 V — the simplest, a analogue voltage sets the dim level, no addressing, no protocol. Widely used with standalone controllers and with greenhouse climate computers. It is also the easiest to get wrong, because a 0-10 V input with a floating reference will drift with cable length, and two fixtures on the same control line can end up at different levels.
DALI — addressable, bidirectional, and the right choice when the room has 200 fixtures and each needs individual scheduling. It costs more per driver and requires a DALI controller, but it gives you per-fixture diagnostics, which matters when a failed fixture is 4 m up.
Proprietary / RJ-style controllers — common in the US grow-light market, where the fixture is sold with a controller that speaks a manufacturer-specific protocol. Convenient, and a lock-in.
Whatever the control method, one requirement is often missed: flicker at low dim levels. A driver dimming to 10% with a 100 Hz output ripple produces a light output that flickers at twice the line frequency, and while plants do not care, cameras do. If the facility does time-lapse monitoring, or the fixture is used in a research or propagation room where imaging is part of the workflow, specify ripple below 5% at 10% dim, or a driver with a higher switching frequency and adequate output capacitance.
Certifications that matter in this market
For the US, the DesignLights Consortium maintains a horticultural qualified products list, and utility rebates are often tied to it. The listing covers the fixture, and the driver’s test data is part of the submission — including the input power at rated output and the power factor. A driver with a PF of 0.6 at 600 W will make the fixture’s power quality figures look bad in a submission that compares against a 0.9 threshold, which is one of the few cases where power factor directly affects whether a rebate is paid.
Safety-wise, UL 8750 for the LED equipment, and UL 1012 if the driver is sold separately. For Canada, CSA C22.2 No. 250.13. For the EU, EN 61347-2-13 with EN 62471 for photobiological safety of the lamp system, which for high-output horticultural fixtures is a real assessment and not a formality — the blue light hazard at close range is significant at these intensities.
What to write into a grow-light driver spec
If you are sourcing for a fixture you will sell on efficiency, the specification needs five lines beyond the wattage:
- Constant power from 60% to 100% of rated output voltage, maintained across ta. Not “constant power capable”.
- Efficiency at 50%, 75% and 100% load, measured at 25 °C case, with the measurement method stated.
- Output ripple at 10% dim, as a percentage of output current.
- Dimming protocol and the control interface, with the addressing scheme if DALI.
- ta and tc, with the tc point marked on the housing.
Those five lines separate a driver that will make your fixture’s datasheet true from one that will make it a conversation with an unhappy customer eighteen months later.
How to verify a constant power claim in an afternoon
Most suppliers will say their driver is constant power. The test that separates a real CP design from a marketing one takes an afternoon and a power meter, and it does not require a photometric lab.
Run the fixture at 25 °C ambient for 30 minutes and record input watts at the wall with a true-RMS meter — a Fluke 1730 or any class-B logger will do. Then run it in a chamber or an insulated enclosure at the declared ta for two hours and record again. Compare the two figures.
A true constant power driver will show a difference under 2%. A constant current driver dressed up as CP will show 3–5%, because the array’s forward voltage fell and the driver did not compensate. A constant voltage driver will show a rise of 3–5%, which is the dangerous direction and the one that accelerates junction temperature.
Then check the window. Set the array to 60% of its rated Vf, if the driver allows it with a variable bench load, and see whether the wattage still holds. If the driver drops into constant current at that point, the CP window is narrower than the datasheet implies, and a cold room or an aged array will push you outside it.
Finally, repeat the 25 °C measurement at 50% dim. This is where a driver that is genuinely 93% efficient at full load falls to 89%, and where the energy arithmetic in the previous section either holds up or quietly disappears.
Two hours of measurement will tell you more about a driver than any datasheet, and it is the difference between selling a fixture on a claim and selling it on a number the customer can reproduce.
FAQ
Why do grow lights need constant power instead of constant current? Because a constant current driver delivers less power as the LED array heats up and its forward voltage falls — typically 3–5% less at operating temperature. Constant power holds the wattage, so the fixture delivers what the label says at 60 °C, not just at 25 °C.
Does constant power make the light brighter? No. It keeps the electrical power constant, which keeps the photon output constant as the array warms. The brightness at cold start is the same; the difference is that it does not drop off.
Is constant voltage ever acceptable for a grow light? Only for low-power arrays with well-controlled thermal design, and even then it is a risk: in CV mode a hot array draws more current, which raises temperature further. On a marginal heatsink this is a thermal runaway loop.
How much does a constant power driver cost compared with constant current? Roughly $2–4 more at 600 W, which is 1–2% of the fixture cost. It pays back in the first efficiency claim you have to defend.
Can I dim a constant power driver to 10%? Most can, but check the output ripple at low dim. Below 20% the loop is operating far from its design point, and some drivers become unstable or produce visible flicker that matters if the room is imaged.
Building a fixture that has to hold its wattage at 60 °C? The constant power LED driver range covers 40 W to 300 W with a defined CP window, and our OEM constant power driver manufacturing line builds the 600 W and 1,000 W versions with the efficiency curve at 50/75/100% included in the test report. Send the array’s Vf range and the ambient profile, and we will size the window before you commit to a housing.